TechTonic Shifts Kluge Hans, the horse who did math, is a lesson for AI

Ever since Apple came with it’s D-Day moment of trying to capture a beachhead on the AI-front, I’ve been using the Think Different campaign intro in different ways. And true to the usual course of events, here’s todays warped version.

This one is for the dreamers who confuse coincidence with brilliance, englightenment in entropy, who find depth in puddles, meaning in madness, and who sees their dog’s wagging tail as a sign of genius.

In this here article, I’ll be talking about Clever Hans (Kluge Hansel in ze German language). Because this is ze horse who convinced an entire nation that he could do math. And the TL;DR of the story is that Hans has a modern day equivalent in the form of Artificial-intelligence-Assistants.

I started in 1905 in Germany. A country that is known for its precision and efficiency, and they brought together an academic panel to conclude that a horse could calculate.

But in the end Hans was not crunching numbers, because he was crunching egos instead.

And there’s a lesson to be learned from the tale of Hans.

Because Hans is what we experience all day when we interact with the likes of ChatGPT, Claude, Meta AI, Googley and whatnot.

Let me explain.

But first, ze commercials:

If you like mein schtuff and vant to support me:

Toss a coin to your witcher, and buy me a coffee ♨️🎄
Connect with me on Linkedin 🙏
Subscribe to TechTonic Shifts to get your daily dose of tech 📰

Meet clever Hans. Ze equestrian “Einstein”

Clever Hans belonged to Wilhelm Von Osten (pronouce “Vilhelm”) .

He was a man with many hats.

He was a math teacher, a horse trainer, a phrenologist (the people that feel your skull and confirm you are mentally insane). Oh yeah, he also was an occasional mystic. Hahahaha… if there was a pseudoscience to dabble in, you bet your ass Vilhelm was there.

But his pièce de résistance was Hans, the so-called calculating horse. Mind you, 1905 was also the Annus Mirabilis of Einstein, the wonder-year, where he published four articles which changed our world for good. So I guess, Vilhelm wanted his 15 minutes of fame as well.

Now, Von Osten claimed that his four-legged friend could add, and subtract, and multiply, and divide, and even solve calendar-based riddles.

For example, if Monday was the 3rd of the month, our Hansel would tap his hoof eight times to answer, “What’s Saturday”

Hans of course, was a parlor trick for Von Osten. Yes, like what Apple claims ChatGPT “reasoning” is: a cheap mix of tricks held together with custom code and ductape. But of course, us sain people who read the papers know that this is just their way of admitting that they are late at the AI party: Apple Intelligence is late to the AI party and brought us… a new set of emoji’s

The duo started touring Germany.

He was showcasing Hans’ “mathematical genius” to the crowds that were packed together on the streets and on fairs. And so Hans was hailed as a wonder of nature, the equine answer to Pythagoras.

The media, like The New York Times, drooled over him like a dog with rabies, and they wrote huge headlines which declared that “a horse had unlocked the mysteries of math”.

So, naturally, Von Osten being the gullible type, he believed that Hans was proof that animals, when they are treated with proper education, could even rival human intellect.

His ambitions were as lofty as they were misguided.

Poor Vilhelm.

Poor Hans.

Hans’ fall from grace

Just like every dramatic tale, after a quick rise to fame, followed by the peak of inflated expectations, inevitably the downfall follows to the trough of disillusionment (sounds familliar, AI?).

Because despite the acclaim, skeptics lingered, and negative press began*

Enter Oskar Pfungst from Frankfurt. And he knew his shit. He was a biologist and psychologist who wasn’t dazzled by Hans’ hoof-tapping theatrics.

Pfungst conducted a lot of rigorous tests on Hans.

He separated Hans from his owner, and asked questions that Von Osten himself didn’t know the answers to, and of course he made sure that Hans couldn’t see the face of the one asking him the questions.

The results were clear: Hans only got the right answer when he could pick up on subtle, involuntary cues from humans, such as slight changes in posture or facial expression when he reached the correct number of taps.

In short, Hans wasn’t doing math.

He was doing what animals do best, and that is reading humans like a book.

This revelation gave birth to the “Clever Hans Effect“, and that is a term that we are going to use from now on, to describe situations where someone (or something) appears to be solving a problem but is really just responding to external cues.

The Kluge Hans effect

Remember the “Who Am I” parroting / mimicking experiment we all did last week?

This experiment led to 62 screendumps and even more inbox messages.

It proved that ChatGPT desperately wants you to believe that it “understands” you, by impersonating your behavior.

Are we shocked by this revelation?

I guess we all saw that one coming.

So clever Hans was the 1905 version of ChatGPT, and in fields like Artificial Intelligence, where systems can seem intelligent they are more often just parroting patterns without true understanding.

So as usual, after fame and fortune comes depression and despair and Hans’ career as a mathematical marvel ended as abruptly as it began.

Of course Von Osten was devastated by the findings, and he withdrew Hans from public life.

Apparently, the horse lived out his remaining days in quiet obscurity. But his legacy lives on in every interaction we have with our Ai-buddies (yes, lower case i).

Hans was not the only animal “mathematician”

Clever Hans wasn’t the only animal to flirt with mathematics, or at least to convince humans he could. Everyone can Google. So I did too. And since I know you wouldn’t, I did it for you:

Here’s number 1:

Jim the wonder dog

Jim was a Setter from Missouri, and he became quite a sensation in the 1930s.

His owner, with the unfortunate name of Sam Van Arsdale (Dutch for: Sam of Assholes), claimed that Jim could answer complex questions such as spelling names or identifying colors, by just pointing to specific cards and objects.

Of course skeptics tested Jim.

They were asking him questions in multiple languages, and that included Morse code, and Jim appeared to be successful.

Sure, he wasn’t a math prodigy per se, but he had this ability to interpret commands, and that earned him a reputation as a genius canine. However, like Hans, many suspected that Jim was simply reading subtle cues from his handlers faces and bodies.

Toto the counting parrot

Toto was an African Grey Parrot, and he stunned audiences in the 1940s with his arithmetic skills.

Toto’s owner was a circus performer, who would ask him questions like, “What’s 3 plus 2”.

And Toto would squawk the answer.

Raaaaaa…

🦜 5.

You may-or-may-not know that parrots are renowned for their mimicry and cognitive abilities.

In other words they can be smart as fuck, but further scrutiny revealed that Toto’s talents were just not about math and more about conditioned responses to – again visual or verbal cues. But nonetheless, he became a favorite among crowds for his performances.

Good for you Toto !

Raaaa! faaak yaa!

Paul the psychic octopus

For the ones not of the TikTok generation (that excludes you, Alexia Fin), we all know the tale of Paul the Octopus. Paul gained fame during the 2010 FIFA World Cup by “predicting” the outcomes of matches.

Paul chose between two boxes, and each had a flag of the competing teams, and his choices were remarkably accurate. He correctly predicted eight matches in a row. There was no numerical reasoning involved, but Paul’s success was statistically improbable and it started a discussion about animal intelligence and intuition.

The math problem that fooled Cambridge

The funny thing about Clever Hans’ story is that it didn’t end with Pfungst’s exposé.

Because in 2002, a Cambridge University’s mathematics entrance exam had a problem that aspiring students needed to solve, which was inspired by Hans. It was about a hypothetical horse named “Harry” who answered math questions by stomping his hoof on a grid. And with each answer corresponding to the number of squares his hoof partially covered. By the wasy, the task was to determine if Harry’s answers were random or deliberate.

It turned out that Harry’s answers were entirely consistent with random chance. His hoof’s placement aligned with probabilities tied to the geometry of the grid. And Like Hans, Harry was a statistical fraud, and it was proven once again that math and horses make for fascinating bedtime stories.

The Kluge Hans story to me is an anecdote to humanity’s eternal hunger for miracles.

The horsy reminds me that we often see what we want to see: A horse doing math or an AI chatbot “thinking” like a human.

And the “Clever Hans Effect” is for all of us who believe that AI will become sentient soon. For the scientists, educators, and technologists that need to question what we observe and also how we interpret it.

So, the next time someone writes about that ChatGPT can predict the stock market, remember Hans and ask yourself: is it genius or just cleverness in disguise?

*Negative press is a clear sign, that a hype is waning.

Signing off from the ruins of dead hopes of AI-sentience,

Marco

Well, that’s a wrap for today. Tomorrow, I’ll have a fresh episode of TechTonic Shifts for you. If you enjoy my writing and want to support my work, feel free to buy me a coffee ♨️

Think a friend would enjoy this too? Share the newsletter and let them join the conversation. Google appreciates your likes by making my articles available to more readers.

Kluge Hans, the horse who did math, is a lesson for AI

If you like mein schtuff and vant to support me: